NUTRICIONISTA JEFE DEL SERVICIO DE ALIMENTACIÓN

Empirical research on GDSS has grown immensely in the past ten years. Laboratory studies have dominated the literature, and they have been the primary method for investigating the effects of GDSS.

Steeb and Johnson (1981) conducted an experiment to compare the effects of a chauffeured GDSS called Group Decision Aid. The system was tested using ten groups each with three members. The decision task was a specialised, complex crisis scenario derived from an international terrorist simulation case. Five groups had computer support while the other five groups relied on manual tools (they were given pencils, paper, and blackboard). All groups were asked to reach a group consensus as to what decision to make. Results indicated that the GDSS groups were more satisfied with the process and more confident with the decision they had made. The GDSS groups also developed higher-quality decisions according to the experts who evaluated the decision outcomes.

Lewis (1982) viewed GDSS research as an intersection of research in decision support systems, social psychology, microcomputer technology, and systems science. He perceived GDSS largely as a way to make formal decision models available to individuals working in groups. Lewis used a microcomputer-based GDSS called ‘Facilitator’ to support three person groups. Much of the research was tocused on evaluating the system design of ‘Facilitator’, and the experiment played an important

role in learning more about obstacles in using a GDSS system. A control treatment group had no support, the first experimental treatment group used ‘Facilitator’, and the second experimental treatment group used a structured paper-and-pencil technique that incorporated the same features as ‘Facilitator’. Lewis found that the use of GDSS produced decisions of higher quality generated more alternatives per decision, and reduced domination by single group members when compared to either the control group or the paper-and-pencil group. This study also provided suggestions for future GDSS design such as informing group members of the time they were taking on different stages of a process and maintaining a record of minutes.

Turoff and Hiltz (1982) conducted two experiments to show that ‘computers may indeed be used to support group communications as an integral part of decision support systems’ (p.83). Their system was built around an automated Delphi technique and provided computer conferencing capabilities. The experiments were conducted to investigate the impact of computed-based support on group decision making. The first experiment was essentially an experiment into the effects of computer conferencing on group decision making. The second experiment, however, manipulated leadership in the group, the use of GDSS and the mode of group interaction (either face-to-face or computer-conference). The results indicated that the use of a GDSS aided the groups in reaching quality decisions more often than groups unaided by a GDSS. The results also revealed that leaders tended to emerge less often in computer-supported groups.

Gallupe (1985) performed an experiment to examine the impact of GDSS on tasks with low and high levels of difficulty. He developed a rudimentary GDSS in BASIC and then used a 2x2 experimental design that compared GDSS and non-GDSS groups performing simple versus complex tasks. Decision outcomes and group processes were measured. Decision outcomes measured included decision quality, decision time, decision confidence, satisfaction with group process, and amount of GDSS usage, the group processes measure consisted of number of issues considered, number of alternatives generated, and participation in decision making. He found that GDSS was particularly appropriate for complex decision tasks. He also found that GDSS groups

made better quality decisions than non-GDSS groups but GDSS groups were less satisfied than non-GDSS groups.

Watson (1987) studied the impact of using a GDSS on consensus formation in small groups. He developed a GDSS in C programming language running under the UNIX operating system. This system became the Software Assisted Meeting Management System (SAMM). In a task requiring resolution of competing personal preferences, 3 or 4 person groups were randomly assigned to one of the three experimental treatments: a computer-based support system; a manual, paper and pencil, support system and no support. He found that GDSS technology appeared to offer some advantage over no support, but little advantage over the paper and pencil method of supporting group discussion. He also found that GDSS groups were less satisfied with the process than non-GDSS groups.

Beauclair (1987) conducted a GDSS study using a small GDSS developed on a small local area network. Groups had to reach a decision regarding a student discipline case. Using a 2x2 factorial design, she compared control groups, groups with computer- aided brainstorming support, groups with computer-aided ranking and voting support, and groups with both computer-aided brainstorming and ranking/voting support. Beauclair found that there were no differences between GDSS groups and non-GDSS groups for decision quality, time to make the decision, amount of participation by group members, and satisfaction with the group outcome.

Jarvenpaa, Rao, and Huber (1988) conducted an experiment in which the same three groups used three different forms of GDSS technology. Groups were given unstructured problems and asked to use either an electronic blackboard, or workstations, or were given no support. They found that GDSS had no effect on participation or satisfaction with the process.

Zigurs, Poole and DeSanctis (1988) reported on a GDSS experiment which focused on group process rather than outcome. Small groups of three people were used. The amount and the pattern of influence behaviour experienced within the group were the

major dependent variables. The major empirical findings of the study showed no significant differences between the overall amount of influence behaviour attempted in computer-supported groups, although significant differences were found in the pattern of influence behaviour, i.e., the different types of influence behaviour used in the computer supported group. They also found that using a GDSS resulted in a more even distribution of influence in the group compared to non-GDSS groups.

Lim, Raman, and Wei (1990) conducted an experiment to look at GDSS effects on the equality of influence and dominance among decision groups. They found no differences in the equality of influence, except for the no-leader GDSS groups, which had a more even distribution of influence. The dominant member in GDSS groups had less influence than the dominant member in non-GDSS groups.

Gallupe (1990) conducted two experiments using two different GDSS but the same type of task to study the effect of use of a GDSS on individual versus group decision making. The study compared the performance of GDSS groups, non-GDSS groups, and the ‘best members’ of those groups to determine if use of a GDSS improved decisions. The results were consistent in both experiments. The findings indicated that GDSS groups did not do as well as the best members of their group and non-GDSS groups did as well as or better than the best members. One explanation proposed was that GDSS generated more equal participation among group members, making it more difficult for the best member to influence the group.

George, et al. (1990) compared the performance of six-member groups, some using a GDSS called GroupSystems and others without GDSS support. Using a generate-and- choose task, they found no differences and sometimes poorer decision quality in the GDSS supported groups. Non-GDSS groups were also more likely to reach consensus in less time but the GDSS supported groups had more equal participation. They also found no significant effects due to anonymity or the presence of a randomly selected leader.

Jessup, Connolly and Galegher (1990) studied the influence of anonymity on group process in groups using GDSS with an idea-generating task. They found that the anonymous GDSS groups generated more comments, were more critical, and were more likely to embellish ideas proposed by others than the non-anonymous groups.

Gallupe and McKeen (1990) conducted a laboratory experiment to examine the effects of the use of a GDSS in face-to-face versus remote meetings. Use of a GDSS was found to increase the time it took to reach a decision but had no impact on the decision quality of the groups. The study also demonstrated that decision satisfaction was less in GDSS groups.

Winniford (1991) studied the use of the GroupSystem GDSS using five and ten member groups for a chosen task. The results were similar to George, Easton, Nunamaker, and Northcraft (1990). There were no differences in decision quality or member satisfaction.

Sengupta and Te’eni (1991) investigated the effect of computer generated cognitive feedback in GDSS supported group decision processes. In a laboratory experiment with groups of three decision makers, 15 groups received on-line cognitive feedback and 15 groups did not. The results showed that users receiving cognitive feedback maintained a higher level of control over the decision making process as their decision strategies converged.

Jessup and Tansik (1991) conducted an experiment using a GDSS to evaluate effects of anonymity and proximity on group process. Twenty groups of four persons each performed an idea-generating task. The results indicated that group members working anonymously and apart generated more comments. Close proximity groups were more satisfied and the highest levels of perceived system effectiveness were reported under the conditions of anonymity.

Cass, Heintz and Kaiser (1992) conducted an experiment to assess the effect of a GDSS on synchronous face-to-face and dispersed meetings with subjects linked via a

voice connection. Members solved a preference allocation task and reported their satisfaction with the meeting process and its outcome. The results demonstrated that both GDSS and location effects were significant for decision quality, with face-to-face non-GDSS group members reporting the highest level of satisfaction. Dispersed subjects without GDSS technology were more satisfied with both the process and the outcome of their meeting. But when a GDSS was available, face-to-face subjects reported higher satisfaction with the meeting outcome than dispersed subjects.

Valacich, Dennis and Nunamaker (1992) studied the effects of group size (3 and 9 members) and group member anonymity on the performance of groups using a computer-mediated idea-generation system. The results indicated that although group members in all conditions made a similar number of comments, larger groups generated significantly more and higher quality ideas than smaller groups. Anonymity however had no effect on group performance. Members of small identified groups made the fewest critical remarks, were the most satisfied, and rated themselves more effective than group members from the other experimental conditions.

Dickson, Partridge and Robinson (1993) studied the effects of facilitative support of GDSS on group performance. The study explored two facilitative supports: chauffeur- driven and facilitator-driven. The results showed that chauffeur-driven groups had higher levels of post-meeting consensus than the facilitator-driven groups; there were no differences with the performance of the person guided groups.

Wheeler, Mennecke and Scudder (1993) reported a laboratory experiment in which they manipulated the degree of restrictive structure in a GDSS and the preference of the group for procedural order. They found that performance was generally better when the group used the non-restrictive GDSS, although they were more satisfied with the restrictive environment. They concluded that preference for procedural order and restrictiveness moderates the manner in which technology-supported decision processes were used and perceived.

Hwang and Guynes (1994) conducted an experiment to investigate the effects of GDSS in groups of nine persons and the effects of group size (three versus nine persons) in the computer supported environment. A 2x2 completely randomised factorial design was employed. The results indicated that decision quality can be improved in large computer-supported groups and large groups generated more alternatives but took longer to reach a final decision than the smaller ones.